Specify what you want it to build, the AI asks for clarification, and then builds it.
GPT Engineer is made to be easy to adapt, extend, and make your agent learn how you want your code to look. It generates an entire codebase based on a prompt.
- Simple to get value
- Flexible and easy to add new own "AI steps". See
steps.py. - Incrementally build towards a user experience of:
- high level prompting
- giving feedback to the AI that it will remember over time
- Fast handovers back and forth between AI and human
- Simplicity, all computation is "resumable" and persisted to the filesystem
Setup:
pip install -r requirements.txtexport OPENAI_API_KEY=[your api key]with a key that has GPT4 access
Run:
- Create a new empty folder with a
main_promptfile (or copy the example foldercp -r example/ my-new-project) - Fill in the
main_promptin your new folder - run
python main.py my-new-project
Results:
- Check the generated files in my-new-project/workspace
Implementing additional chain of thought prompting, e.g. Reflexion, should be able to make it more reliable and not miss requested functionality in the main prompt.
Contributors welcome! If you are unsure what to add, check out the ideas listed in the Projects tab in the GitHub repo.
You can specify the "identity" of the AI agent by editing the files in the identity folder.
Editing the identity, and evolving the main_prompt, is currently how you make the agent remember things between projects.
Each step in steps.py will have its communication history with GPT4 stored in the logs folder, and can be rerun with scripts/rerun_edited_message_logs.py.